Details
-
Improvement
-
Status: Closed
-
Major
-
Resolution: Fixed
-
None
Description
We found out that calculateSharedOffHeapNodeMemory is called too many times during scheduling (from java profiling result). Add caching in Cluster.calculateSharedOffHeapNodeMemory should speed up the scheduling process
Attachments
Issue Links
- requires
-
STORM-3595 Refactor Resource Aware Strategies (Base, Generic, Default)
- Resolved
- links to